summree
How to use Codex Better than 99% of People
OpenAI
Jack Roberts

How to use Codex Better than 99% of People

3 min read6 May 2026Worth watching
TL;DR
Jack Roberts provides a comprehensive walkthrough of OpenAI's Codex app, showing how it combines coding, chat, and autonomous agents into one platform powered by GPT-4.5. The video covers skills, plugins, automations, memory systems, and a multi-model strategy using Claude and Gemini inside Codex to maximize output quality.
Key points
1
Codex is powered by GPT-4.5 and uses ~4x fewer output tokens per task than Claude, making it more cost-efficient for heavy users
2
Skills (like Claude's skills) can be created, saved, and even migrated from Claude to Codex using a prompt-based export workflow
3
Automations let you schedule agents to run on timers without supervision — functioning like Claude Code but with a built-in scheduler
4
The three-brain strategy tags in Claude for design work and Gemini for long video analysis directly inside Codex via the terminal
5
Chronicle (Pro Max, $200/month) captures screen context via screenshots so Codex has live awareness of what you are working on
Actionable insights
Start on the $20 ChatGPT Plus plan and only upgrade to Pro ($100) or Pro Max ($200) when you consistently hit usage limits
Export all Claude skills using a migration prompt, then re-import them into Codex to avoid rebuilding workflows from scratch
Build UIs in Lovable first for best design output, sync to GitHub, then clone the repo into Codex for further development
Use the terminal inside Codex to run Claude Code instances alongside ChatGPT, letting each model handle its strength — Claude for design, Gemini for long-form video analysis
Set in-chat reminders and standalone automations for recurring tasks like daily Git summaries, bug watches, or database refactors
Notable quotes

Codex uses around four times fewer output tokens per task. So even if you said that Opus was more powerful, it is four times as much work to get to the same outcome.

My philosophy is that the best way forward is what I call the multi-brain strategy — leveraging the best models for the particular use case in hand.

It will not store what you were ordering controversially at 11 p.m. on Valentine's Day. It will delete that as it goes.

Worth watching?
Worth watching the full video?
Watch if you want the live demo footage of Codex in action — the key workflows are all captured here, but seeing the UI, Chronicle, and multi-model terminal setup visually will help it click faster.
Topics
AI & TechOpenAI

Explore more summaries on these topics →

Saved you some time? The creator still deserves a like.

Watch on YouTube →
More like this

Want this for your own channels?

Add the channels you follow. Every new video summarised and in your inbox the moment it drops. From £4/month.

Try it free